Why MENA mobile growth needed its own platform
A lot of growth tools work everywhere. Madar AI exists because the ones that matter most for subscription apps don’t.
I want to be specific about that, because “we’re built for the region” is the kind of marketing copy that doesn’t survive a serious technical review. So here’s the actual list — six years of running MENA growth on tools built for somewhere else, and the things that quietly cost real money.
1. Your LTV model thinks Ramadan is noise
Almost every off-the-shelf LTV model treats sharp engagement spikes as outliers to smooth over. In MENA, that means it smooths over the most important month of your year.
Ramadan engagement on a meditation app in Saudi Arabia isn’t a 1.2x bump. It’s 2-3x for the right cohort, sometimes 4x. Strip that out and your LTV forecast for the Riyadh acquisition cohort is suddenly 30-40% too low. You under-bid. Your competitor with the same product but the right model wins the auction. You lose users you should have won.
We built a model that understands Ramadan as a season, not as anomaly. Same logic for Hajj for utilities apps, Eid for shopping apps, the Turkish back-to-school cycle for kids’ content.
2. Your attribution stack disagrees and nobody tells you
Pull AppsFlyer, Singular, Meta, Google, and App Store Connect for the same yesterday. You will get five different numbers for “new paid subscribers from Meta acquisition.” Not slightly different — meaningfully different.
The US growth playbook is “pick a source of truth and ignore the rest.” That works when the gap is 3%. In MENA, where iOS adoption skews high in GCC and the iOS attribution graph is half-broken post-ATT, the gap is regularly 15-25%. Pick wrong and your CAC math is fiction.
Madar AI shows you all sources side by side, surfaces the delta, and tells you which one to trust for which decision. (Hint: ad spend reconciliation uses Meta. Cohort retention uses Singular. Revenue uses RevenueCat. There is no single source of truth and pretending otherwise is how you lose 18 months of compounding budget.)
3. Your benchmarks come from someone else’s product
When a US growth tool tells you “good trial-to-paid conversion is 8-12%”, it’s quoting a benchmark from US-tier-1 apps. Apply that bar to a fitness app in Tunisia and you’ll either think you’re failing (you’re not) or you’ll set a target that requires a 4x improvement that no app of your category has ever achieved in your market.
Real benchmarks are category × country × monetisation model. We have those for MENA because we built the dataset ourselves. Nobody else was going to.
4. Your payment stack is invisible to the analytics
App Store and Play Store are 70% of MENA subscription revenue in most categories. iyzico, Papara, Tabby, and Tamara take the rest — and that share is growing fast as KSA, UAE, and Turkey push local payment rails.
Most analytics tools treat the App/Play subscription as canonical and silently lose visibility on the rest. So your “retention curve” is actually two retention curves stapled together — App Store users with full visibility, and local-payment users where you can see the first purchase and basically nothing after.
We pull all of them. iyzico’s settlement webhook, Papara’s reconciliation file, the Stripe billing for the Turkish lira flow — all of it shows up in the same cohort retention chart with the source labelled.
5. Your creative testing is monolingual
If you run a Turkish fitness app, your creatives need Turkish copy that doesn’t sound machine-translated, Arabic creatives for the GCC expansion, and English creatives for the affluent expat segment. That’s three creative pipelines minimum.
Every US growth tool I’ve used treats creative performance as a single English-language stream. You get one performance number per creative. There is no way to ask “which Arabic headline outperforms which Turkish headline for the same offer.” We make that the default view, not a hidden filter.
6. The recommendation is “spend more on Meta”
I’ll keep this one short. Generic ad-platform recommendation engines have one core suggestion: spend more on the channel that’s working. This is fine until your TikTok is saturated in Saudi Arabia at 6% CTR and your Snapchat is wide open at 2% CTR but converts 3x. The right answer is “rebalance” and no off-the-shelf tool will tell you that without a custom integration.
Madar AI’s recommendation engine is built around budget reallocation across channels, not channel-by-channel optimisation in isolation. That’s the difference.
What we’re not yet
I want to be honest about scope. Madar AI is not a replacement for AppsFlyer or RevenueCat or your ad platforms. We sit on top of them. We make them tell the truth.
We’re also not a panacea. If your product has a retention problem rooted in the core experience, no growth platform fixes that — and we’ll tell you when the diagnosis is “this is a product problem, not a growth problem.” That’s a feature.
If you’ve felt any of the six points above, I’d love to compare notes. The product page is the marketing pitch, but the conversation that matters happens by email.
— Mehmet